Plugin (or adapter) for VXG Server to perform face recognition using AWS Rekognition service.
Copyright (c) VXG 2019
Smart camera sends an event with image to VXG Server when it detects that probably there's some face on the image.
This module gets these events along with images and send them to AWS Rekognition to find a known faces there.
If there's no known faces, it adds new faces to collection at AWS Rekognition to recognise them in the future.
Face identifiers (UUIDs) and rectangles are sent back to the VXG Server allowing users to process this meta, for example search the particular face in the VXG Server' storage.
The Big picture:
| This module |
| | <=> AWS Rekognition
Smart camera => VXG Server <= | PollingImageSource => Queue => Worker |
| | => VXG Server
PollingImageSource periodically polls VXG Server for images that cameras marked as having some face. It gets storage
URL to the images and passes them to processing Queue. Those events are marked with "processing" meta tag.
Worker gets image URLs from the Queue and starts it's processing:
1) downloads image from the storage;
2) sends "search_faces_by_image" request to AWS Rekognition, get face UUID if someone familiar is found
3) if no known faces is found, but there's a faces, call "index_faces" to add faces to Collection, get their UUIDs
4) set metadata with face UUID and rectangle to this event, also set meta tag "processed_has_face"
5) if no faces found set meta tag "processed_no_face"